Laying Waste to Mercury: Inexpensive Sorbents Made from Sulfur and Recycled Cooking Oils

Abstract Mercury pollution threatens the environment and human health across the globe. This neurotoxic substance is encountered in artisanal gold mining, coal combustion, oil and gas refining, waste incineration, chloralkali plant operation, metallurgy, and areas of agriculture in which mercury‐rich fungicides are used. Thousands of tonnes of mercury are emitted annually through these activities. With the Minamata Convention on Mercury entering force this year, increasing regulation of mercury pollution is imminent. It is therefore critical to provide inexpensive and scalable mercury sorbents. The research herein addresses this need by introducing low‐cost mercury sorbents made solely from sulfur and unsaturated cooking oils. A porous version of the polymer was prepared by simply synthesising the polymer in the presence of a sodium chloride porogen. The resulting material is a rubber that captures liquid mercury metal, mercury vapour, inorganic mercury bound to organic matter, and highly toxic alkylmercury compounds. Mercury removal from air, water and soil was demonstrated. Because sulfur is a by‐product of petroleum refining and spent cooking oils from the food industry are suitable starting materials, these mercury‐capturing polymers can be synthesised entirely from waste and supplied on multi‐kilogram scales. This study is therefore an advance in waste valorisation and environmental chemistry.


Table of Contents Page Number
General Experimental Considerations S3 Synthesis of canola oil polysulfide (50% sulfur) S5 Synthesis of polysulfide from olive oil and sunflower oil S6 Lipid analysis of oils S7 Synthesis of canola oil polysulfide with varying sulfur content S12 Synthesis of canola oil polysulfide using recycled cooking oil S13 IR analysis of canola oil polysulfide S14 NMR analysis of CDCl 3 soluble fraction of non-porous canola oil polysulfide (50% sulfur) S15 Reaction of sulfur and fatty acid methyl esters derived from plant oils S16 SEM analysis of canola oil polysulfide (50% sulfur) S18 EDS analysis of canola oil polysulfide (50% sulfur) S19 Auger analysis of canola oil polysulfide (50% sulfur) S20 Raman spectroscopic analysis of canola oil polysulfide (50% sulfur) S21 Ellman's test on the canola oil polysulfide S23 TGA of canola oil polysulfide (30, 50, 60 and 70% sulfur) S24 Simultaneous thermal analysis (DSC and TGA) of unreacted vegetable oils S25 DSC of non-porous canola oil polysulfides and free sulfur analysis S27 Comparison of polysulfides prepared from canola oil, olive oil, and sunflower oil (TGA and DSC) S30 Comparison of canola oil polysulfide to commercial factice (IR, Raman, TGA, DSC) S32 Comparison of canola oil polysulfide prepared by inverse and classical vulcanisation (TGA and DSC) S35 Dynamic mechanical analysis of canola oil polysulfide S37 Glass transition temperature by DSC for non-porous canola oil polysulfide (50% sulfur) S38 Mercury chloride capture from water (including SEM and EDS analysis of polymer) S39 Leaching of mercury(II) from canola oil polysulfide S43 Mercury metal capture from water (including SEM, EDS, and Auger analysis of polymer) S44 XPS analysis of canola oil polysulfide before and after mercury capture S47 XRD analysis of canola oil polysulfide before and after mercury metal capture S48 Mercury capture using a polysulfide prepared from recycled cooking oil S49 Mercury capture using commercial factice F17 S49 Chromogenic response of canola oil polysulfide after reaction with mercury metal S50 Mercury flour preparation and SEM and EDS analysis S51 Capturing mercury flour with the non-porous canola oil polysulfide S55 Toxicity studies S57 Synthesis of a porous canola oil polysulfide (50% sulfur) using a sodium chloride porogen S59 SEM analysis of porous canola oil polysulfide (50% sulfur) S60 Raman spectroscopic analysis of porous canola oil polysulfide (50% sulfur) S61 TGA and DSC of porous canola oil polysulfide (50% sulfur) S62 Glass transition temperature by DSC for porous canola oil polysulfide (50% sulfur) S63 NMR analysis of CDCl 3 soluble fraction of porous canola oil polysulfide (50% sulfur) S64 Mercury vapour experiments using the porous canola oil polysulfide S65 Installation of thiols on the porous canola oil polysulfide by partial reduction with NaBH 4 S66 Experiments on mercury bound to natural organic matter (NOM) S67 Sulfate release from the porous canola oil polysulfide S69 Experiments using an organomercury fungicide ] S70 References S72 Simplified structure of the canola oil and the canola oil polysulfide. Note that the polysulfide can potentially crosslink inter-and intramolecularly.

Polysulfide prepared from sunflower oil and from olive oil
Sunflower and olive oil polysulfides were prepared using the same procedure as to prepare canola oil polysulfide. Sulfur (20.0 g) was added to a 250 mL round bottom flask and heated, with stirring, to 180 °C. After 5 minutes of heating at this temperature the sulfur turned from a yellow to an orange liquid. At this point, the sunflower or olive oil (20.0 g) was added dropwise over 5 minutes. After 12 minutes, the reaction with sunflower oil reached its gel point and formed a rubber. The reaction with the olive oil reached its gel point after 21 minutes of reaction time. Both samples were left to cool for 15 minutes before removing from their flasks. A third reaction prepared with canola oil was carried out for comparison. All samples were independently washed by submerging in 0.1 M aqueous NaOH for 90 minutes followed by washing with DI water and drying in open air. The samples have the same physical appearance:

Fig S2 |
A polysulfide rubber is obtained by the reaction of an equal mass of sulfur and olive oil, sunflower oil, or canola oil. The time to reach the gel point is shorter for sunflower oil, likely because of its higher polyunsaturated linoleic acid content in the triglyceride.

Lipid analysis of vegetable oils
Vegetable oil (1.00 g) was mixed with methanol (100 mL) in a 250 mL round bottom flask and cooled to 0 °C. Sodium methoxide (100 mg) was then added to the stirred mixture. The reaction mixture was stoppered and stirred vigorously at room temperature for 24 hours. Vigorous stirring is important to ensure effective mixing of the two phases present at the start of the reaction. After 24 hours, the reaction was cooled to 0 °C and quenched with 0.1 M HCl (10 mL). The mixture was transferred to a separatory funnel and then diluted with ethyl acetate (150 mL) and water (150 mL). The organic layer was isolated and then washed with water (3 x 50 mL) and brine (3 x 50 mL) before drying (sodium sulfate), filtering and concentrating under reduced pressure. Analysis by 1 H NMR and GC-MS revealed clean conversion to the fatty acid methyl esters. Typical yields for fatty acid methyl esters from 1.00 g vegetable oil: Canola oil: 800 mg Sunflower oil: 800 mg Olive oil: 780 mg General assignments for the 1 H NMR for the mixture of fatty acid methyl esters. Note that the relative integration will depend on the degree of unsaturation.

Method for GC-MS analysis
Fatty acid methyl esters prepared from the oils as described in the previous experiment were prepared as a solution in chloroform (~5 mg/mL) and then analysed by GC-MS using the following method on a Varian CP3800: Hold at 50 ºC for 1 min, ramp from 50 to 200 ºC at a rate of 25 ºC /min. Slow ramp to 3 ºC/min rate from 200 to 230 ºC and hold at 230 ºC for 25 min. Next ramp from 230 ºC to 280 ºC at 25 ºC /min and hold at 280 ºC for 10 min. The total run time: 54 minutes. Injection temp: 250 ºC, carrier gas flow rate 1.2 mL/min. Representative GC traces are shown below with the major fatty acid methyl esters labelled. Methyl ester molecular ions were identified by comparison to the major fragmentation product, [M-31] + , due to a loss of the methoxy group.

Preparing the Canola Oil Polysulfide at Different Sulfur Compositions
Canola oil polysulfides were prepared with different sulfur content by varying the ratio of canola oil to sulfur used in the synthesis. In a typical synthesis, sulfur was heated to 180 ºC and the corresponding mass of sulfur was added slowly to maintain a constant internal temperature. All reactions were carried out on a 40 g scale. The two phase mixture was stirred rapidly to ensure efficient mixing. Typically, all samples reached the gel point within 20 minutes. Even prolonged heating (50 minutes) of the reaction mixture containing 10% sulfur did not result in a rubber.

Fig. S6
| The reaction of canola oil with sulfur at different mass ratios. At 10% sulfur, a liquid was obtained. Between 20% and 70% sulfur, the product was a rubber. At 80% sulfur and higher, the product was brittle. c.

Synthesis of polysulfide using recycled cooking oil
Used cooking oil was obtained from a local café after it had been used to fry various foods for one day. The oil was used as received and was not purified in any way. In the synthesis of the polysulfide, sulfur (10.0 g) was added to a 250 mL round bottom flask and heated, with stirring, to 180 ºC. 5 minutes after reaching this temperature, the sulfur turned from a yellow to orange liquid. At this point, the crude, recycled cooking oil (10.0 g) was added dropwise to the sulfur over a period of 5 minutes. After 22 minutes of additional reaction time, the mixture reached its gel point and formed a brown rubber. The polymer was removed from the flask with a metal spatula. The product was washed by submerging the polymer chunks in 0.1 M NaOH for 90 minutes, followed by washing with water. Air drying provided the final product.

Fig. S7
| Photograph of a polysulfide prepared from used cooking oil (top). IR spectra of the polysulfide prepared from unused canola oil and the polysulfide prepared from waste cooking oil (both 50% sulfur).

NMR analysis of CDCl 3 soluble fraction of non-porous canola oil polysulfide (50% sulfur).
CDCl 3 (10 mL) was added to a sample of non-porous canola oil polysulfide (500 mg). The mixture was stirred vigorously at room temperature for 15 minutes to extract the soluble fraction of the polymer. The resulting solution was filtered and then analysed by 1 H NMR. The spectra are shown below with unreacted canola oil as a reference. The key findings from this experiment are that the soluble fraction of the polymer contains unreacted alkene peaks. However, even in this soluble fraction the ratio of the terminal methyl groups of the fatty acid ester (0.87 ppm) and the alkene peaks (5.25-5.37 ppm) have changed. In the canola oil this ratio is 1.0 : 1.0. and in the soluble fractions of the polymer it is ~3:1, indicated alkene reaction. After NMR analysis, the solvent was evaporated to determine the amount of polymer dissolved. For the non-porous canola oil polysulfide, 61 mg dissolved (12% of the polymer mass). From this result, it can be concluded that in the polymerisation of canola oil and sulfur at a mass ratio of 1:1, the gel point is reached before all alkenes are consumed.
1 H NMR of canola oil (before polymerisation with sulfur) 1 H NMR of CDCl 3 soluble fraction of non-porous canola oil polysulfide

Reaction of sulfur and fatty acid methyl esters derived directly from vegetable oils
Sulfur and fatty acid methyl ester obtained from canola oil. Sulfur (87 mg, 0.34 mmol S 8 ) was added to a 100 mL round bottom flask and then heated to 180 ºC with stirring. The methyl ester prepared from transesterification of canola oil with sodium methoxide (Fig S3) (100 mg) was then added to the sulfur. The reaction was stirred at 180 ºC for 30 minutes and then cooled to room temperature to provide a viscous black oil. The mixture was analysed directly by 1 H NMR. All alkene peaks (5.0-5.5 ppm) were consumed in the reaction: Sulfur and fatty acid methyl ester obtained from sunflower oil. Sulfur (404 mg, 1.56 mmol S 8 ) was added to a 100 mL round bottom flask and then heated to 180 ºC with stirring. The methyl ester prepared from transesterification of sunflower oil with sodium methoxide (Fig S3) (500 mg) was then added to the sulfur. The reaction was stirred at 180 ºC for 30 minutes and then cooled to room temperature to provide a viscous black oil. The mixture was analysed directly by 1 H NMR. All alkene peaks (5.0-5.5 ppm) were consumed in the reaction: Sulfur and fatty acid methyl ester obtained from olive oil. Sulfur (440 mg, 1.72 mmol S 8 ) was added to a 100 mL round bottom flask and then heated to 180 ºC with stirring. The methyl ester prepared from transesterification of olive oil with sodium methoxide (Fig S3) (500 mg) was then added to the sulfur. The reaction was stirred at 180 ºC for 30 minutes and then cooled to room temperature to provide a viscous black oil. The mixture was analysed directly by 1 H NMR. All alkene peaks (5.0-5.5 ppm) were consumed in the reaction:  To confirm that sulfur reacts at the alkenes present in the vegetable oils, the reaction between elemental sulfur and the methyl ester derived from each oil was studied. S18

SEM analysis of Canola Oil Polysulfide (50% sulfur)
The Canola Oil Polysulfide was prepared according to the standard procedure, providing a distribution of particles from 0.2 to 12 mm. These particles were then passed through two polyethylene sieves to obtain particles in the range of 0.5 to 1 mm. A sample of these particles were then mounted on an aluminium SEM pin mount using carbon tape before sputter coating with platinum. Subsequent SEM analysis revealed the surface of the polysulfide to be microtextured-a property that increases surface area.
Canola Oil Polysulfide (before exposure to mercury):

Scanning Auger Electron Spectromicroscopy of canola oil polysulfide (50% sulfur)
The non-conductive nature of the samples meant that for a useful Auger Electron Spectrum to be obtained, a 2 nm layer of Platinum was needed to provide conductivity to the surface of the sample. The elemental maps of carbon and sulfur show that the carbon-sulphur ratio varies spatially.

Fig. S13
| Auger spectroscopy of the canola oil polysulfide (50% sulfur) revealed strong signals for carbon and sulfur, consistent with the proposed structure. Raman analysis shows stretches at 432 cm -1 and 470 cm -1 , consistent with S-S vibrational modes of a polysulfide material. Peaks at 1437 cm -1 and 2900 cm -1 are attributed to the canola oil domain of the polymer. The Raman spectra for the canola oil polysulfide (50% sulfur), and the canola oil and sulfur starting materials are shown below: Fig. S15 | Raman spectra of the canola oil polysulfide (50% sulfur) and the canola oil and sulfur starting materials.

Confocal Raman images of Canola Oil Polysulfide (50% sulfur)
Confocal Raman images were acquired for the Canola oil polysulfide and are displayed in Figure S16. Figure  S16a is an optical image of the sample with figures S16b and c representing confocal Raman images (30x30 µm) of exactly the same area of the sample. Figures S16d and e are zoomed in Raman images (15x15 µm) of the same area with the centre of each image corresponding to the white and black crosses in figures S16 b and c. The data in figures S16b and d were generated by plotting the intensity of the 470 cm -1 region of each Raman spectrum while the data in figures S16c and e were generated by plotting the intensity of the 2900 cm -1 region of each Raman spectrum. The Raman spectra that are present in the brighter regions of figures S16b and d typically have the appearance of the sulfur starting material displayed in figure S15 (orange curve) and the Raman spectra that are present in the brighter regions of figure S16b and e typically have the appearance of the canola oil polysulfide copolymer (50% sulfur) also displayed in figure S15 (green curve). It is apparent from figure S16b that there are regions of free sulphur embedded in the polysulfide matrix that form what appear to be small microparticles (5 to 15 µm in size). This data supports the SEM/EDS analysis as well as other results recently reported in the literature on related composites. 1

Analysis of thiol-content on the canola oil polysulfide surface using Ellman's test
A sample of canola oil polysulfide (1.00 g, 50% sulfur) was placed into each of three 50 mL centrifuge tubes along with 8 mL phosphate buffer (100 mM, pH 8) and Ellman's reagent (8 mg, 0.020 mmol). As a control, Ellman's reagent was also added to three separate samples of buffer in the same way, except in the absence of polymer. All samples were mixed on a lab rotisserie for 1 hour at room temperature before filtering. The filtrates were then diluted 7-fold and analysed by UV-Vis spectroscopy. Absorbance at 412 nm are listed below. No reaction with Ellman's reagent was observed, as no significant increase in absorbance at 412 nm was observed (student t-test). Therefore, thiol content on the polymer is negligible and consistent with the proposed polysulfide structure.

Thermogravimetric Analysis of the Canola Oil Polysulfide at Different Sulfur Compositions
TGA of the canola oil polysulfide was carried out for the canola oil polysulfide prepared at 30, 50, 60 and 70% sulfur by weight. The first major mass loss at ~250 ºC increased in proportion to the amount of sulfur in the polymer. We therefore attribute the first mass loss to thermal degradation of the polysulfide domain of the polymer. Consistent with this interpretation, the end of the first mass loss of each polymer (400 ºC) corresponds well with the mass of sulfur in each polymer (30% mass loss for the 30% sulfur polysulfide , 50% mass loss for the 50% sulfur polysulfide, 60% mass loss for the 60% sulfur polysulfide and 70% mass loss for the 70% sulfur polysulfide). The second mass loss occurs upon decomposition of the canola oil domain of the polymer.

Estimation of free sulfur in canola oil polysulfides
Quantitative DSC was used to determine free sulfur content in the Canola Oil Polysulfides. S 8 has a distinctive DSC peak at 125 °C that stretches from 100 °C to 150 °C. The area of this peak (from 100 °C to 150 °C) increases linearly with sulfur mass. On average 1 mg sulfur gave a response of 49.3 J/g within the range tested. This response was used to approximate the free sulfur present in the polymer. Because the free sulfur may be present in forms other than S 8 , this is only an estimate. The calibration curve is show below:

Comparison of Canola Oil Polysulfide and Factice
Factice is a commercially available additive used extensively in the rubber industry. Factice is made through classic vulcanisation of vegetable oils, such as canola oil. Typically, this involves adding low percentages of sulfur to hot vegetable oil, resulting in cross-linking of the oil. In contrast, the canola oil polysulfide reported in this manuscript is prepared by inverse vulcanisation where the vegetable oil is added to high mass percentages of liquid sulfur, thereby crosslinking the polysulfide. Because both factice and the canola oil polysulfide are made with similar starting materials, we were interested to compare the two materials directly (spectroscopically, thermally and in its binding to mercury). Shown below, along side the canola oil polysulfide, are photographs of factice samples with 10%, 17% and 25% sulfur. These samples were generously provided by D.O.G. Chemie.

Raman Spectra of factice and canola oil polysulfide:
Raman spectra were obtained for F10 and F25 Factice and compared to the canola oil polysulfide (50 wt% sulfur) prepared by inverse vulcanisation. The increased sulfur content results in an increased intensity of peaks at 432 and 470 cm -1 . This is consistent with greater polysulfide (S-[S] n -S) content in the 50 wt% canola oil polysulfide in comparison to F10 or F25 Factice

S35
Comparison of canola oil polysulfide prepared by inverse vulcanisation and classic vulcanisation. The canola oil polysulfide was prepared with 50% sulfur according to the standard inverse vulcanisation procedure (Fig. S1). For classic vulcanisation, canola oil (10.0 g) was heated to 180 ºC in a 250 mL round bottom flask with stirring. Sulfur (10.0 g) was then added in several portions over 5 minutes. The mixture was stirred vigorously for an additional 15 minutes, after which time the mixture reached its gel point and formed a brown rubber very similar in appearance to the product formed from inverse vulcanisation. STA of both samples revealed a similar decomposition and calorimetric profile.

DSC of canola oil polysulfide prepared by traditional vulcanisation and inverse vulcanisation
Differential scanning calorimetry was repeated, with a focus on the region where sulfur melts. Slightly more free sulfur was observed when using inverse vulcanisation (9% free sulfur) compared to traditional vulcanisation (8% free sulfur).

Determination of glass transition temperature by DSC for non-porous polysulfide (50% sulfur)
The glass transition temperature of the non-porous canola oil polysulfide was -12.2 ºC, as determined by DSC:

Canola Oil Polysulfide capture of mercury chloride from water
The Canola Oil Polysulfide (2.0 g, mixture of particles 2-12 mm in diameter) was added to a 20 mL glass vial, followed by 5 mL of a 20 mg/mL aqueous HgCl 2 solution (100 mg total HgCl 2 ). The mixture was incubated without stirring for 24 hours. A control sample containing just water and the polysulfide (and no HgCl 2 ) was also run in parallel. After the 24 hours, the polysulfide was isolated by filtration and washed with 3 aliquots of 5 mL deionised water. The aqueous solution was then transferred to a pre-weighed 50 mL round bottom flask and the water removed by rotary evaporation to provide unsequestered HgCl 2 . The experiment was run in triplicate resulting in an average of 46 mg of HgCl 2 remaining in solution and 54 mg bound to the polysulfide. Notably, the polysulfide underwent a change in colour during the incubation, from brown to grey. No colour change was observed if mercury was not present. The material changed colour (from brown to grey) after binding the mercury.
Effect of the amount of Canola Oil Polysulfide on the capture of mercury chloride from water. The procedure above was repeated with different quantities of Canola Oil Polysulfide: 250 mg, 500 mg, 1.00 g, 2.00 g, 4.00 g and 8.00 g. The volume and concentration of aqueous HgCl 2 remained the same for each sample (5 mL of a 20 mg/mL aqueous solution of HgCl 2 ), as did the incubation time (24 hours). As the mass of polysulfide increases, the mass of HgCl 2 remaining in solution after the 24 hour incubation decreases. This is likely because of the increased surface area available to bind to mercury. This experiment also indicates that the maximum amount of mercury chloride bound by weight for this particle size is about 4%.

SEM analysis of Canola Oil Polysulfide after treatment with mercury chloride
A 12.0 g sample of the Canola Oil Polysulfide (0.5 to 1.0 mm particles, as prepared above using sieves) was incubated in an aqueous solution of mercury chloride (30 mL of 20 mg/mL HgCl 2 ) for 24 hours. After this time, the polysulfide turned from brown to grey. The polysulfide was then filtered and washed with deionised water (3 × ~30 mL). The filtrate was concentrated under reduced pressure to provide 186 mg of unbound mercury chloride. Therefore, the polysulfide had captured 414 mg (or 70%) of the mercury. A sample of the mercury-treated polysulfide was then prepared for SEM and analysed.  Mercury rich nanoparticles were detected on the surface of the polymer at Spot 1. The unmodified canola oil polysulfide is detected at Spot 2. A sulfurrich particle was detected in Spot 3.

Mercury Leaching Study (mercury chloride)
1.0 g samples of mercury chloride-treated polysulfides were incubated in 10 mL milliQ water for 24 hours (2.2 mg total HgCl 2 ). The water was then tested by ICP-MS against an ICP standard of Hg in 2% HNO 3 (1% HNO 3 and 1% HCl in water used as a diluent) to determine the concentration of mercury that had leached from the polymer over this time. Tests were run in duplicate. Both samples were diluted 1/10 in a 1% HNO 3 and 1% HCl in water matrix. Samples were run in He mode to ensure ions flew monatomically. This result indicates that leaching into water is negligible. If all mercury chloride were leached from the polymer, a concentration of 0.22 mg/mL or 220,000 ppb would be measured. An average of only 0.57 ppb Hg 2+ was detected in the leachate.

Canola Oil Polysulfide reactive capture of liquid mercury metal [Hg (0) ]
The Canola Oil Polysulfide (1.00 g, mixture of particles 2-12 mm in diameter) and 100 mg elemental mercury were added to a glass vial containing 7 mL deionised water. The mixture was stirred vigorously for 24 hours at room temperature. After this time, no elemental mercury was visible and the polysulfide had changed colour from brown to black. The colour change occurred after approximately four hours of vigorous stirring at room temperature. The colour change correlates with mercury capture and occurs on the surface of the particle. After the 24 hours of stirring, the black polymer-bound mercury was isolated by filtration and dried to constant mass. A mass of 1.099 g of this material was isolated, indicating good mass balance in the mercury capture (e.g. > 99% of the mercury reacted with the polysulfide). Left: Canola Oil Polysulfide Right: Reaction product of the polysulfide with mercury metal. Left: Particle of the polysulfide after reaction with mercury metal Right: Severed particle reveals that mercury is bound only to surface of particle. S45

EDS analysis of elemental mercury-treated polysulfide surface
The Canola Oil Polysulfide (50% sulfur) and elemental mercury were added to a glass vial containing 7 mL DI water and reacted as described previously. After the reaction (24 hours, vigorous stirring), no elemental mercury was visible and the polysulfide had changed colour from brown to black. The polysulfide was isolated by filtration and then a 10 mm particle was cut in half. The cross-section was profiled by SEM and EDS, revealing the mercury was bound only to the surface, where the material appeared black.  eV XRD Sample Preparation 1.24 g elemental mercury was added to a 50 mL centrifuge tube containing 2.47 g sulfur and mixed for 24 hours using an end-over-end mixer. Similarly, 2.47 g of canola oil polysulfide (50% sulfur, < 0.5 mm particle size) was mixed with 1.52 g elemental mercury in an end-over-end mixer for 24 hours. Unreacted sulfur, unreacted polysulfide, as well as those samples reacted with elemental mercury, were all ground to a fine powder using a mortar and pestle in preparation for loading on an XRD sample stage. The XRD spectra obtained for both reactions was metacinnabar, as it was identical to previously published XRD spectra. 2 It can therefore be concluded that the black material that results from the reaction of mercury metal and the S-S bonds of the canola oil polysulfide is metacinnabar. Fig S45 | XRD scans of a, elemental sulfur, b,

d. Metacinnabar formed by reaction of polysulfide and mercury metal
Mercury capture using polysulfide prepared from recycled cooking oil 1.0 g of the polysulfide (50% sulfur) prepared from recycled cooking oil (Fig. S7) was placed in a 25 mL round bottom flask equipped with a stirrer bar, along with elemental mercury (171 mg) and 10 mL DI water. The flask was sealed and the mixture stirred for 24 hours. During this time the polysulfide turned black, and some unreacted elemental mercury was still visible. The polymer and mercury were separated by mixing with equal volumes of hexane and water. The polymer remained at the phase boundary and the mercury settled to the bottom of the aqueous phase. The water and mercury were isolated, and separated from the polymer. The mercury was then separated from the water by transferring to a separatory funnel and diluting with dichloromethane. The mercury-dichloromethane mixture was then isolated and the dichloromethane evaporated in a fume hood. The mass of the unreacted mercury was recorded.

Mercury capture using Factice F17 (D.O.G.)
2.8 g of F17 grad D.O.G. Factice was placed in a 25 mL round bottom flask equipped with a stirrer bar, along with elemental mercury (217 mg) and 10 mL DI water. The flask was sealed and the mixture stirred for 24 hours. During this time the factice darkened in colour, and some unreacted elemental mercury was still visible. The factice and unreacted mercury were separated by mixing with equal volumes of hexane and water. The polymer remained at the phase boundary and the mercury settled to the bottom of the aqueous phase. The water and mercury were isolated, and separated from the polymer. The mercury was then separated from the water by transferring to a separatory funnel and diluting with dichloromethane. The mercury-dichloromethane mixture was then isolated and the dichloromethane evaporated in a fume hood. The mass of the unreacted mercury was recorded. Fig. S46 | Factice F17 (17% sulfur) and a polysulfide prepared from recycled cooking oil (50% sulfur) were compared in their reaction with mercury metal. An amount of polymer was added such that the mass of sulfur was the same. Both samples captured virtually the same amount of mercury metal, suggesting that the amount of mercury that can react corresponds to the amount of sulfur in the polysulfide. This result also suggests that the polysulfides in factice can react with mercury metal and that free sulfur is not required.

Sensitivity of chromogenic response of canola oil polysulfide in its reaction with mercury metal
In order to test the sensitivity of the polysulfide's response to elemental mercury, quantities of mercury ranging from 72 to 285 mg were added to 10 and 20 g quantities of polysulfide in separate 50 mL centrifuge tubes (Fig. S47). The polymer-mercury mixtures were rotated on a lab rotisserie for 24 hours and any changes to the mixture recorded. In all cases the polymer turned black, indicating reaction of mercury with the polysulfide. Given the intensity of the colour change, it is presumed that the polymer may also turn black when exposed to lesser quantities of elemental mercury than shown here. Because of the difficulties in measuring small quantities of metallic mercury, this experiment was not pursued further. From these results we can conclude that mercury can be detected by visual inspection after the reaction of mercury and the canola oil polysulfide at ratios of 3.6 mg of mercury per gram of polymer or lower.

Mercury flour preparation and SEM and EDS analysis
Loam was obtained from Glenalta, South Australia and ground with a mortar and pestle before passing through a sieve to obtain a soil with particle size less than 0.50 mm. 5.00 g of this powdered soil was sealed in a 50 mL centrifuge tube with 200 mg elemental mercury and rotated (30 rpm) on an end-over-end mixer for 24 hours. After this time the mercury was no longer visible to the naked eye, having been dispersed throughout the soil. There was no visible difference between the soil before or after treatment with mercury.

Soil
Soil milled with mercury metal: "Simulated Mercury Flour" Loam, prepared at a particle sizes < 0.50 mm Soil milled with mercury appeared highly similar to the untreated soil  After thorough searching, mercury was detected as microspheres dispersed in the soil. This "floured mercury" is covered in micro-and nanoparticles of soil. The soil prevents the mercury from coalescing.

EDS Spot 1
Microparticle of mercury, coated in nanoscale soil particles

EDS Spot 2
Soil particle, adsorbed to the surface of a mercury microparticle

S55
Capturing mercury flour using the non-porous canola oil polysulfide 5.0 g canola oil polysulfide (50% sulfur) of a particle range of 2.5 -5.0 mm was isolated using a sieve. These particles were added to 5.0 g of the simulated mercury flour and mixed in a 50 mL centrifuge tube on an endover-end mixer for 24 hours. A control sample treated identically but without the addition of mercury was also prepared for comparison. Over this time, the polymer in the presence of mercury turned black, indicating reaction with the mercury flour. The polymer in the soil in which no mercury was added remained brown. The polymer particles were then separated from the bulk of the soil using a sieve. EDS analysis clearly indicated that mercury was bound to the polymer. This experiment demonstrates that the canola oil polysulfide, prepared as a particle, can capture mercury from soil and then be isolated using a sieve.
The isolated polymer after incubation with mercury-treated soil: After mixing with the mercury flour, the polymer changed from brown to black, indicating reaction with the elemental mercury.
The isolated polymer from the control sample (soil, but no mercury). After mixing with just soil, the polymer remains its original brown colour The top image is the black polymer, isolated from the soil using a sieve. The black colour is consistent with reaction with the mercury flour. In a control experiment (bottom), the polymer retains its brown colour after mixing with soil that does not contain mercury.

Control polymer (isolated from soil with no mercury)
Polymer, post treatment (isolated after reaction with mercury flour) Polymer mixed with soil, separated using a sieve and then washed with 3 x 10 aliquots DI water to remove some of the soil Polymer mixed with mercury flour, separated using a sieve and then washed with 3 x 10 aliquots DI water to remove some of the soil particles
Cytotoxicity of mercury-treated and untreated polysulfides in HepG2 and Huh7 cells. Cells were cultivated as described above and seeded in 24 well-Transwell® plates at a concentration of 30 000 cells/well (300 µl), and allowed to adhere to the bottom of the well for 24 h. At this point, culture medium was removed and 200 µl of fresh complete medium was added to the bottom layer. Also, 3.75 mg or 37.5 mg of treated or untreated polysulfide was added to each insert in technical duplicates, and 100 µl of complete medium was added on top of the polysulfide, thus creating a continuous layer of medium on top of the cells and the polysulfides. Cells were incubated for another 22 h 30, at which time cell viability was assessed as described above. Results are shown as average of 3 independent experiments (bars), and error bars represent standard error of the mean. There was no difference in cell viability for the cells treated with polymer and cells treated with the polymerbound mercury. Under these conditions, neither the polymer nor the polymer-bound mercury exhibit significant toxicity.
Notes for Figs S54 and S55: Dose 1 = 37.5 mg of polymer in 300 mL of culture medium Dose 2 = 3.75 mg of polymer in 300 mL of culture medium  Cytotoxicity and estimation of IC 50 of HgCl 2 in HepG2 and Huh7 cells. Cytotoxicity of HgCl 2 was assessed using a CellTiter-Blue ® Cell Viability Assay (Promega, USA), a fluorescent dye approach based on the ability of metabolically active cells to convert the dye resazurin to the fluorescent resorufin product. Briefly, cells were seeded at a concentration of 10 000 cells/well (100 µL) in flat-bottom 96 well-plates and allowed to adhere and adapt to the plates for 24 h. At this point, culture medium was exchanged to complete medium supplemented with increasing concentrations of HgCl 2 in technical triplicates (1, 5, 10, 30, 60, 80, 100 µM). Plates were incubated for 22 h 30 min, at which time cell viability was assessed by exchanging the culture medium to medium supplemented with CellTiter-Blue Reagent (dilution 1:20 from commercial stock) and incubated for another 1 h 30 min, before analysis of fluorescence on an Infinite M200 (Tecan, USA) plate-reader (λ exc =530, λ em =590). Relative fluorescence units (R.L.U.) were normalized to the values obtained for the appropriate vehicle controls. Results are shown as average of 3 independent experiments. A sigmoidal curve (variable slope) was fitted to each dataset, using GraphPad Prism v5 software, and used to calculate the half maximal inhibitory concentration (IC 50 ) of HgCl 2 on both cell-lines. The average IC 50 was 40 µM for HepG2 cells and 34 µM for Huh7 cells.

Synthesis of a porous canola oil polysulfide (50% sulfur) using a sodium chloride porogen
Sulfur (3.00 g) was added to a 250 ml round bottom flask equipped with a 40 mm oval stirring bar. The flask was then placed in a DrySyn reactor cup preheated to 180 °C. The mixture was stirred slowly as the sulfur melted and turned into an orange liquid. At this time, canola oil (food grade, 3.00 g) was added dropwise over 2 minutes to maintain a temperature near 180 °C. After the addition of the canola oil, sodium chloride (14.00 g, previously powdered using a mortar and pestle) was added in several portions over 10 minutes. The addition of the sodium chloride results in a thick, paste-like mixture. The rate of stirring was adjusted to ensure steady mixing. Typically 15-20 minutes after the addition of sodium chloride was complete, the reaction mixture vitrifies and turns into a hard brown solid. After vitrification, the flask was removed from the DrySyn heater and allowed to cool for 1 hour. The product was removed from the flask using a spatula and then milled for 1 minute in a blender (8.5 cm rotating blade). The resulting material (20.0 g) was then transferred to a beaker, followed by 150 mL of deionised water. The mixture was stirred for 1 hour at room temperature to leach the sodium chloride from the polymer particles. The particles were isolated by filtration and then washed a second time in the same manner to ensure the complete removal of sodium chloride. Isolating the polymer by filtration and then drying under high vacuum provided the final porous polymer as a sponge-like material (6.0 g). If residual sodium chloride is observed on the surface of the polymer or by SEM, additional water washes can be used. The amount of free sulfur in the porous canola oil polysulfide was estimated to be 13% by mass, as determined by integration of the peak from 100 ºC to 150 ºC (see Fig 23 for the analogous experiment for the non-porous polysulfide and calibration curve).

Determination of glass transition temperature by DSC for porous polysulfide (50% sulfur)
The glass transition temperature of the porous canola oil polysulfide was -12.9 ºC, as determined by DSC:

NMR analysis of CDCl 3 soluble fraction of porous canola oil polysulfide (50% sulfur).
CDCl 3 (10 mL) was added to a sample of the porous polysulfide (508 mg). The mixture was stirred vigorously at room temperature for 15 minutes to extract the soluble fraction of the polymer. The resulting solution was filtered and then analysed by 1 H NMR. The spectra are shown below with unreacted canola oil as a reference.
The key findings from this experiment are that the soluble fraction of the polymer contains unreacted alkene peaks. However, even in this soluble fraction the ratio of the terminal methyl groups of the fatty acid ester (0.87 ppm) and the alkene peaks (5.25-5.37 ppm) have changed. In the canola oil this ratio is 1.0 : 1.0. and in the soluble fractions of the polymer it is ~3:1, indicated alkene reaction. After NMR analysis, the solvent was evaporated to determine the amount of polymer dissolved. For the porous canola oil polysulfide, 111 mg dissolved (22% polymer mass). From this result, it can be concluded that in the polymerisation of canola oil and sulfur at a mass ratio of 1:1, the gel point is reached before all alkenes are consumed.
1 H NMR of canola oil (before polymerisation with sulfur) 1 H NMR of CDCl 3 soluble fraction of porous canola oil polysulfide

Mercury vapour experiments using the porous canola oil polysulfide
Hg 0 removal tests were performed using a fixed bed-reactor as shown in Fig. S63. The inlet Hg 0 vapour was generated using a mercury permeation device (VICI metronics), which was operated at 60 °C. The porous canola oil polysulfide (300 mg) was placed in the quartz glass reactor (1 cm internal diameter), occupying a volume of approximately 0.4 cm 2 . N 2 gas with a flow rate of 0.1 L/ min, which contained 586.4 µg/Nm 3 Hg 0 , was introduced to the reactor using mass flow controllers. At this volume of sorbent and flow rate, the residence time is 0.24 seconds-a challenging test for the polysulfide sorbent. All elemental and oxidised mercury exiting the reactor were measured quantitatively using a modified Ontario Hydro Method (OHM), in which KCl (0.01 M) and KMnO 4 /H 2 SO 4 (20 mg L −1 ) impinger solutions were used in the train of traps as mercury absorbing media. Elemental mercury (Hg 0 ) is captured by the KMnO 4 solution, whereas any oxidised mercury (Hg 2+ ) is trapped by the KCl solution. The remaining adsorbed mercury was retained on the canola oil polysulfide. Cold vapour atomic fluorescence spectroscopy (CV-AFS) was used to measure the collected Hg from the system after the Hg 0 removal experiments. In all experiments, the amount of oxidised mercury (Hg 2+ ) collected from the KCl traps was negligible (<< 1% of total Hg). Hg 0 removal efficiency of material was determined by the following equation: The effect of operating temperature on mercury removal efficiencies of the developed material was tested by varying the reactor temperature from 25-100 °C. It was hypothesised that the rate of reaction between the mercury vapour and the polysulfide would increase with temperature-a requirement for continuous processes S66 with short residence times such as those in this experiment. It was found that the material had highest Hg 0 removal efficiency of 66.5 % at 75 °C.

Fig. S64
| 75 ºC was found to be an optimal temperature for capturing mercury in a continuous process, with 66.5% of the mercury removed from the gas stream over a residence time of approximately 0.24 seconds.

Installation of thiols on the porous canola oil polysulfide by partial reduction with NaBH 4
The porous canola oil polysulfide (2.00 g) was added to a 100 mL round bottom flask along with 34 mg sodium borohydride. Methanol (10 mL) was added carefully and the reaction mixture was stirred open to air for 1 hour. After this time the reaction was quenched with 10% HCl (10 mL) and then diluted further with 10 mL H 2 O. The resulting product was isolated by filtration and dried under vacuum. This material was positive in an Ellman's test in which 10 mg of the Ellman's reagent was added to a sample of the polymer in 10 mL phosphate buffer (100 mM, pH 8.0), indicating the presence of thiols on the surface of the polymer. The partially reduced polymer was very similar in appearance to the original porous polysulfide. Using a larger excess of sodium borohydride (>500 mg) led to substantial degredation and loss of porous and particulate character, consistent with reduction of the polysulfide.

Materials and Methods
Mercury speciation can significantly affect reactivity of mercury and its interaction with sorbent materials. The speciation of mercury in aquatic ecosystems is typically dominated by association with natural organic matter (NOM). Suwannee River aquatic natural organic matter (SR-NOM), reference material 2R101N (International Humic Substance Society) and a 1 ppm Hg(NO 3 ) 2 standard (Brooks Rand Instruments, Seattle, WA, USA) were used to prepare Hg-NOM complexes containing 40 µg/L Hg and 2400 µg/L total carbon (C NOM ) equivalent to a molar Hg:C NOM ratio of 1.8·10 -5 . SR-NOM was dissolved in 10 mM sodium phosphate buffer (pH 7.8) and filtered through a 0.2 µm syringe filter to remove residual particulates. Hg(NO 3 ) 2 was added and the pH was re-adjusted to 7.8 and allowed to age at 4 ºC for at least 5 days. The Hg-NOM stock solution was diluted with 10 mM sodium phosphate buffer to obtain working solutions with Hg concentrations from 0.2 to 7.7 µg/L. A dilution series of the 1 ppm Hg(NO 3 ) 2 standard in 10 mM sodium phosphate buffer was prepared as an NOM-free control.
Where Y is the concentration of the adsorbate on the sorbent, Y max is the sorption capacity, C eq is the solution concentration at equilibrium and K L is the Langmuir adsorption equilibrium constant. The isotherm fits for all COP variants are shown in Fig. S66. The results show that all tested COP samples removed >90% of Hg when added as Hg(NO 3 ) 2 . The strong complexation of mercury with functional groups on NOM competes with the sorption of Hg to any sorbent, thus presenting a unique challenge for the removal of Hg from contaminated ecosystems. Under the conditions of the isotherm experiments, a dilution series was prepared from a concentrated Hg-NOM stock solution. Thus, the concentration of Hg is coupled the concentration of NOM. In a freshwater creek ecosystem, the level of NOM can span a wide range of concentrations, while the level of Hg typically corresponds to the low end of the experimental range, even in contaminated systems. 3 Efficient removal of Hg from solutions containing strong Hg-NOM complexes is achievable as it is determined by the sorbent to solution ratio and the concentration of Hg-NOM. A measure of how efficient the sorbent can remove the contaminant at a specific concentration can be obtained as follows: Where R is the removal efficiency, C 0 is the initial Hg concentration and C eq is the Hg concentration after equilibration with the sorbent. Surface modification of canola polysulfide had a significant impact on Hg removal, with the higher surface area of the porous versions significantly improving removal efficiency. At the lowest initial Hg-NOM concentrations (0.2 µg/L Hg) and a sorbent to solution ratio of 1/300, R was 36% for non-porous canola oil polysulfide, 79% for porous canola oil polysulfide, and 81% for the partially reduced porous polysulfide. The results show that the surface modification of COP, particularly the increased surface area in porous COP, results in a highly effective sorbent which can sorb Hg in the presence of competing ligands such as NOM.

Fig. S66
| Equilibrium sorption data (dots) and fits to isotherm models for the sorption of Hg at low mercury concentrations. 95% confidence bands are shown in gray. A. Unmodified COP with Hg added as Hg(NO 3 ) 2 and linear fit (blue), residual standard error of the fit: 0.21 µg/g; B. Unmodified COP with Hg added as Hg-NOM complex and model fit to the Langmuir isotherm model (red). Langmuir fit parameters: K L = 1.35 L/µg, Y max = 0.21 µg/g, residual standard error of the fit: 0.032 µg/g; C. Porous COP with Hg added as Hg(NO 3 ) 2 and linear fit (blue), residual standard error of the fit: 0.71 µg/g; D. Porous COP with Hg added as Hg-NOM complex and model fit to the Langmuir isotherm model (red). Langmuir fit parameters: K L = 0.46 L/µg, Y max = 1.11 µg/g, residual standard error of the fit: 0.061 µg/g; E. Partially reduced porous COP with Hg added as Hg(NO 3 ) 2 and linear fit (blue), residual standard error of the fit: 0.65 µg/g; F. Partially reduced porous COP with Hg added as Hg-NOM complex and model fit to the Langmuir isotherm model (red). Langmuir fit parameters: K L = 1.29 L/µg, Y max = 0.44 µg/g, residual standard error of the fit: 0.065 µg/g Sulfate release from the porous canola oil polysulfide High sulfate concentrations in low oxygen subsurface environments can result in increased production of methylmercury. Sulfate-reducing bacteria have been associated with mercury methylation and are considered the primary methylators in marine and estuarine environments. 4,5 We therefore determined sulfate concentrations in solutions obtained from batch sorption studies (see Fig. S67). Briefly, 30 mL Hg-NOM complex dissolved in 10 mM sodium phosphate buffer (pH 7.8) at various concentrations were added to amber glass vials containing approximately 100 mg of canola oil polysulfide and equilibrated for 48 hours on a rotary shaker. The solid to solution ratio was constant for all samples. Sulfate concentrations were determined by ion chromatography with a Dionex ICS 2100 AS9HC9 (Dionex Instruments Corporation, Sunnyvale, CA, USA) from filtered sample solutions using 9 mM K 2 CO 3 as the eluent. The amount of sulfate released was normalized to the mass of the polysulfide for each sample. The amount of sulfate released correlated with the concentration of Hg-NOM initially added to the sample (Fig. S67). In the absence NOM, sulfate concentrations were typically <100 µg per g of sorbent. For samples containing NOM, the sulfate concentration was proportional to the NOM concentration.

Mercury removal from a MEMC solution using columns prepared from soil and the porous canol oil polysulfide
Four types of columns were prepared (in triplicate) in the barrel of a 10 mL syringe. The plunger was first removed and cotton wool was used to plug the outlet. The column was then packed with one of the following: soil (3.0 g); soil (1.5 g) and porous polysulfide (1.5 g) mixed together; soil (1.5 g) layered on top of a layer of porous polysulfide (1.5 g), separated by cotton; or porous polysulfide (3.0 g). A solution of MEMC was prepared at 0.15 g/L and then 3 mL of this solution was added to the column by pipette. The plunger of the syringe was carefully re-inserted and the solution was eluted slowly by applying gentle pressure. The total elution time was approximately 2.5 minutes for each type of column. The flowthrough was collected and sample of each was diluted 100,000-fold in a 2% HNO 3 matrix and Hg content was measured by ICP-MS as described in the previous experiment. The mercury concentration for the MEMC solution before passing through the column was also measured in triplicate. The columns and data are shown below. Soil alone (3.0 g) retained 46% of the mercury; soil and polymer (1.5 g each) mixed randomly together retained 66% of the mercury; soil (1.5 g) layered on top of the polymer (1.5 g) retained 75% of the mercury; and polymer alone (3.0 g) retained 73% of the mercury.